in vivo selection of hematopoietic progenitor cells and...

Research article

1952 TheJournalofClinicalInvestigation http://www.jci.org Volume 119 Number 7 July 2009

In vivo selection of hematopoietic progenitor cells and temozolomide dose intensification

in rhesus macaques through lentiviral transduction with a drug resistance geneAndre Larochelle,1 Uimook Choi,2 Yan Shou,3 Nora Naumann,2 Natalia A. Loktionova,4

Joshua R. Clevenger,1 Allen Krouse,1 Mark Metzger,1 Robert E. Donahue,1 Elizabeth Kang,2 Clinton Stewart,5 Derek Persons,3 Harry L. Malech,2 Cynthia E. Dunbar,1 and Brian P. Sorrentino3

1National Heart, Lung and Blood Institute (NHLBI) and 2National Institute of Allergy and Infectious Diseases (NIAID), NIH, Bethesda, Maryland, USA. 3Division of Experimental Hematology, Department of Hematology, St. Jude Children’s Research Hospital, Memphis, Tennessee, USA. 4Department of Cellular and Molecular Physiology, Pennsylvania State University College of Medicine, M.S. Hershey Medical Center,

Hershey, Pennsylvania, USA. 5Department of Pharmaceutical Sciences, St. Jude Children’s Research Hospital, Memphis, Tennessee, USA.

MajorlimitationstogenetherapyusingHSCsarelowgenetransferefficiencyandtheinabilityofmosttherapeuticgenestoconferaselectiveadvantageonthegene-correctedcells.Oneapproachtoenrichforgene-modifiedcellsinvivoistoincludeintheretroviralvectoradrugresistancegene,suchastheP140KmutantoftheDNArepairenzymeO6-methylguanine-DNAmethyltransferase(MGMT*).Wetransplanted5rhesusmacaqueswithCD34+cellstransducedwithlentiviralvectorsencodingMGMT*andafluorescentmarker,withorwithouthomeoboxB4(HOXB4),apotentstemcellself-renewalgene.Transgeneexpres-sionandcommonintegrationsitesinlymphoidandmyeloidlineagesseveralmonthsaftertransplanta-tionconfirmedtransductionoflong-termrepopulatingHSCs.However,allanimalsshowedonlyatran-sientincreaseingene-markedlymphoidandmyeloidcellsafterO6-benzylguanine(BG)andtemozolomide(TMZ)administration.In1animal,cellstransducedwithMGMT*lentiviralvectorswereprotectedandexpandedaftermultiplecoursesofBG/TMZ,providingasubstantialincreaseinthemaximumtolerat-eddoseofTMZ.Additionalcyclesofchemotherapyusing1,3-bis-(2-chloroethyl)-1-nitrosourea(BCNU)resultedinsimilarincreasesingenemarkinglevels,butcausedhighlevelsofnonhematopoietictoxicity.InclusionofHOXB4intheMGMT*vectorsresultedinnosubstantialincreaseingenemarkingorHSCamplificationafterchemotherapytreatment.OurdatathereforesuggestthatlentivirallymediatedgenetransferintransplantedHSCscanprovideinvivochemoprotectionofprogenitorcells,althoughselectionoflong-termrepopulatingHSCswasnotseen.

IntroductionRetroviral vector–mediated gene transfer into HSCs has the poten-tial to improve treatment for various genetic, malignant, and infec-tious diseases (1, 2). However, a critical limitation to stem cell gene therapy is the low gene transfer efficiency in large animal models and clinical trials (3, 4).

In a recent gene therapy trial for X-linked SCID, although engraftment of transduced HSCs was low, as evidenced by the low gene marking in myeloid cells, patients demonstrated near-complete repopulation with genetically modified T lymphocytes (5). In vivo selective survival advantage of lymphocytes carrying the therapeutic transgene (the IL-2 common γ–receptor chain,

encoded by Il2rg) contributed substantially to clinically benefi-cial levels of genetically corrected lymphocytes after gene therapy. These results strongly suggest that providing an in vivo selective advantage to transduced cells may be an attractive method of enhancing the repopulation of transduced cells in vivo. However, for most blood disorders, there will be no constitutive selective advantage of the gene-corrected cells.

One potential approach to in vivo selection of transduced HSCs is to incorporate drug resistance genes to convey protection against chemotherapeutic agents (6). Successful in vivo selection was pre-viously documented in murine HSC transplantation models using several drug resistance genes, including multidrug resistance pro-tein–1 (MDR-1; refs. 7, 8), dihydrofolate reductase (DHFR; ref. 9), and O6-methylguanine-DNA methyltransferase (MGMT; refs. 10–12). However, in more clinically relevant large-animal models, in vivo selection using DHFR (13) and MDR-1 (14, 15) was vari-able and transient, with levels of gene-modified cells returning to baseline within a few weeks. MGMT is one of the most promis-ing drug resistance genes that encodes for the DNA repair protein O6-alkylguanine-DNA-alkyl-transferase (AGT). This protein confers resistance to the cytotoxic effects of alkylating agents, including nitrosoureas [e.g. 1,3-bis-(2-chloroethyl)-1-nitrosourea (BCNU)] and temozolomide (TMZ). To allow hematopoietic

Authorshipnote: Andre Larochelle, Uimook Choi, and Yan Shou contributed equally to this work.

Conflictofinterest: The authors have declared that no conflict of interest exists.

Nonstandardabbreviationsused: ANC, absolute neutrophil count; BCNU, 1,3-bis-(2-chloroethyl)-1-nitrosourea; BG, O6-benzylguanine; CAG, cancer-associated gene; HOXB4, homeobox B4; IS, integration site; LAM-PCR, linear amplification–mediated PCR; LTR, long-terminal repeat; MGMT, O6-methylguanine-DNA methyltransferase; MGMT*, MGMT P140K; MSCV, murine stem cell virus; P2A, 2A peptide from por-cine teschovirus–1; PB, peripheral blood; T2A, 2A peptide from thosea asigna virus; TMZ, temozolomide; YFP, yellow fluorescent protein.

Citationforthisarticle: J. Clin. Invest. 119:1952–1963 (2009). doi:10.1172/JCI37506.

research article

TheJournalofClinicalInvestigation http://www.jci.org Volume 119 Number 7 July 2009 1953

selection, O6-benzylguanine (BG) is coadministered with TMZ for inactivation of endogenous MGMT activity. Several mutant forms of MGMT that give rise to altered AGT are highly resistant to inactivation by BG, but retain their ability to repair DNA dam-age (16, 17). The most commonly used mutant is MGMT P140K (MGMT*), which is almost completely resistant to BG, yet has a modest reduction in the rate of DNA adduct repair compared with wild-type AGT. This mutant form of MGMT can be incorporated in the gene transfer retroviral vectors to facilitate a greater selective advantage for primitive hematopoietic cells containing the vector. In vivo selection of HSCs using MGMT* was successfully achieved in murine (10–12) and canine models (18–20) and in human NOD/SCID-repopulating cells (21, 22). The same approach has also been proposed as a means to protect patients against dose-limiting hematopoietic toxicity following administration of che-motherapy combining alkylating agents and BG for the treatment of brain tumors (23). Protection of the BM compartment would allow administration of increased doses of chemotherapy agents to effect an improved therapeutic response.

A second approach to achieving therapeutic levels of gene-cor-rected cells is to confer a selective growth advantage upon gene-transduced cells. Homeobox B4 (HOXB4) is a potent stem cell self-renewal gene. In murine transplantation models, overexpression of HOXB4 has previously been shown to induce ex vivo expansion and self renewal of HSCs without compromising differentiation or homeostatic regulation of HSC pool size in murine transplanta-tion models (24–26). More recently, HOXB4 function was exam-ined in a clinically relevant nonhuman primate transplantation model: HOXB4 overexpression in rhesus CD34+ cells was able to expand short-term repopulating cells, but had a less pronounced effect on long-term repopulating cells (27).

Drug resistance gene–mediated in vivo selection can amplify the percentage of gene-transduced HSCs; however, it is accompanied by a decrease in the overall number of HSCs. In murine transplan-tation studies, this limitation can be overcome by coexpressing HOXB4 with the drug resistance gene to allow selective regenera-tion of transduced cells after treatment with cytotoxic drugs (28). In this study, we examined the effects of an MGMT*-mediated drug-resistant system with or without an HOXB4-mediated HSC self-renewal system in transplanted rhesus macaques. Our findings indicate that in vivo chemoprotection of progenitor cells occurred, but selection of long-term repopulating HSCs was not seen.

ResultsCD34+ cells transduced with MGMT* lentiviral vectors show transient chemoprotection after in vivo treatment with BG and TMZ. We inves-tigated whether in vivo chemoprotection could be achieved after transduction with MGMT*-expressing lentiviral vectors based on SIV and autologous transplantation of mobilized peripheral blood (PB) or BM CD34+ cells in 5 rhesus macaques (Table 1). Gene transfer efficiency in the CD34+ cell population before transplant in rhesus macaques varied between 18% and 53%, as determined by flow cytometry (Table 1).

In a first set of experiments, AMD3100-mobilized PB CD34+ cells from animals RQ4876 and RQ4795 were transduced using SIV vectors expressing GFP-MGMT* fusion protein (Figure 1A). We administered 2 daily doses of 500 cGy total body irradiation before infusion of 5.6–6.4 × 106 CD34+ cells/kg. Hematopoietic recovery was documented between 7 and 17 days (Table 1). At 3–6 months after transplant, each animal received the first of 4 successive BG/

Tab

le 1

Reco

nstit

utio

n of

irra

diat

ed rh

esus

mac

aque

s w

ith tr

ansd

uced

, aut

olog

ous

CD34

+ cel

ls

Anim

al n

o.

CD34

+ cel

l sou

rce

Lent

ivira

l vec

torA

CD34

+ cel

l In

fusi

on d

ose

ANC

exce

edin

g In

viv

o tre

atm

ent (

no. c

ycle

s)

In v

ivo

Toxi

city

tra

nsdu

ctio

n (c

ells

/kg)

50

0/μ

l (d)

BG

/TM

ZB BG

/BCN

UC se

lect

ionD

MGM

T* v

ecto

rRQ

4876

AM

D310

0-m

obili

zed

PB a

pher

esis

GF

P-M

GMT*

50

%

6.4 ×

106

7 4

3 Tr

ansi

ent

Deat

h, p

ulm

onar

yRQ

4795

AM

D310

0-m

obili

zed

PB a

pher

esis

GF

P-M

GMT*

27

%

5.6 ×

106

17

4 0

Tran

sien

t Al

ive

RQ45

13

G+S-

mob

ilize

d PB

aph

eres

is

MGM

T*-Y

FP

24%

6.

7 ×

106

11

2 2

Tran

sien

t De

ath,

pul

mon

ary

GIRQ

4099

G+

S-m

obili

zed

PB a

pher

esis

M

GMT*

-YFP

18

%

8.5 ×

106

11

1 0

Tran

sien

t De

ath,

radi

atio

n in

duce

dRQ

4152

G+

S-m

obili

zed

BM

MGM

T*-G

FP

53%

2.

1 ×

106

8 7

1 Tr

ansi

ent

Deat

h, G

IM

GMT*

-HOX

B4-G

FP v

ecto

rRQ

4513

G+

S-m

obili

zed

PB a

pher

esis

M

GMT*

-HOX

B4-G

FP

20%

6.

7 ×

106

11

2 2

Tran

sien

t De

ath,

pul

mon

ary

GIRQ

4099

G+

S-m

obili

zed

PB a

pher

esis

M

GMT*

-HOX

B4-G

FP

15%

6.

7 ×

106

11

1 0

Tran

sien

t De

ath,

radi

atio

n in

duce

d RQ

4152

G+

S-m

obili

zed

BM

MGM

T*-H

OXB4

-GFP

50

%

1.6 ×

106

8 7

1 Tr

ansi

ent

Deat

h, G

I

Ani

mal

s R

Q48

76 a

nd R

Q47

95 r

ecei

ved

cells

tran

sduc

ed w

ith M

GM

T*

vect

ors;

ani

mal

s R

Q45

13, R

Q40

99, a

nd R

Q41

52 r

ecei

ved

cells

tran

sduc

ed w

ith M

GM

T*

or M

GM

T*-

HO

XB

4-G

FP

vec

tors

in a

com

peti-

tive

repo

pula

tion

setti

ng (

outc

ome

data

are

rep

eate

d fo

r th

e sa

me

anim

al).

AA

ll le

ntiv

iral v

ecto

rs w

ere

amph

otro

pic

and

had

SIV

env

elop

e. B

For

RQ

4876

and

RQ

4795

, BG

and

TM

Z c

once

ntra

tions

wer

e 12

0 an

d 24

5 m

g/m

2 , r

espe

ctiv

ely.

For

RQ

4513

, RQ

4099

, and

RQ

4152

, BG

con

cent

ratio

n w

as 1

20 m

g/m

2 an

d T

MZ

con

cent

ratio

n w

as in

esc

alat

ing

dose

s: 1

70, 2

02, 2

30, 2

84, 3

25, 3

65, a

nd 4

50 m

g/m

2 . C

For

all

anim

als,

BG

and

BC

NU

con

cent

ratio

ns w

ere

120

and

20 m

g/m

2 , r

espe

ctiv

ely.

DIn

clud

es b

oth

lym

phoi

d an

d m

yelo

id s

elec

tion.

research article


TMZ treatments given at 1- to 2-month intervals. Substantial myelosuppression was seen in both animals after drug adminis-tration (Figure 2A).

Animal RQ4876 showed a rapid increase in transduced PB cells of all lineages in vivo after drug treatment. Steady-state marking lev-els of 2%, 6%, and 5% in granulocytes, lymphocytes, and monocytes, respectively, increased to peak levels of 52%, 33%, and 36% in the respective lineages after the first (granulocytes), second (lympho-cytes), or third (monocytes) dose of BG/TMZ. The fourth dose of BG/TMZ led to only moderate increases in GFP+ cells in all lineages. Treatment with BG/TMZ was well tolerated, with little to no cumu-lative toxicity. The proportion of GFP+ cells subsequently declined gradually over a period of 200 days after the last BG/TMZ admin-istration; marking levels in PB stabilized to values similar to those observed at steady state before the first treatment with BG/TMZ, with 2% GFP+ granulocytes, 6% GFP+ lymphocytes, and 3% GFP+ monocytes (Figure 2B). Lentiviral vector provirus copy numbers were also determined by real-time PCR at several time points in DNA from PB (data not shown). Overall, there was a good cor-relation between flow cytometry and real-time PCR results, which suggests that on average, repopulating cells contained no more than 1 vector copy and that gene silencing was not responsible for the gradual decrease in gene marking levels in vivo.

Animal RQ4795 had very low steady-state gene marking levels in PB granulocytes (0.1%), lymphocytes (0.01%), and monocytes (0.2%; Figure 2B). We tested the safety and feasibility of improving very low baseline gene marking levels with in vivo infusion of BG/TMZ. No change in GFP+ granulocytes was detected at any time point after BG/TMZ administration. A peak increase in GFP+ lymphocytes

(0.2%) and monocytes (1.5%) was observed after the second dose of BG/TMZ, but the increase was only transient, returning to baseline levels within 30 days. The third and fourth doses of BG/TMZ resulted in substantial myelosuppres-sion, but had no impact on gene mark-ing levels in any lineage.

In a second set of experiments, we tested whether alternative SIV MGMT*-expressing vectors in combination with minor modifications to transduction conditions and escalating doses of TMZ could result in improved longer-term gene marking levels. Rhesus competi-tive repopulation experiments were set up using equal numbers of mobilized PB or BM CD34+ cells transduced with SIV vectors expressing MGMT* alone (containing GFP or yellow fluorescent protein [YFP]) or MGMT*-HOXB4-GFP (see Methods and Figure 1). Mul-ticistronic cassettes containing GFP or YFP were generated in each vector using the 2A peptide from porcine teschovi-rus–1 (P2A) and thosea asigna virus (T2A) (Figure 1, B–D). All 3 animals in this study — RQ4513, RQ4099, and RQ4152 — were mobilized with G-CSF plus SCF and received 2- to 3-fold more CD34+ cells/kg compared with the

AMD3100-mobilized animals RQ4876 and RQ4795 after total body irradiation conditioning (Table 1). Hematopoietic reconsti-tution was rapid, with the absolute neutrophil count (ANC) reach-ing 500 cells/μl or more by 8–11 days after transplant. All animals showed a total of 3%–6% marked PB granulocytes and monocytes, and 1%–12% marked PB lymphocytes, at 60 days after transplant (Figures 3, 4, 5). After stable engraftment, we evaluated the drug combination BG/TMZ. The dose of BG was fixed at 120 mg/m2, but, in contrast to animals RQ4876 and RQ4795, the dose of TMZ was gradually escalated from 170 to 450 mg/m2/d for a 5-day peri-od. All animals showed significant myelosuppression after each treatment (Supplemental Figure 1; supplemental material avail-able online with this article; doi:10.1172/JCI37506DS1).

The first animal in this group, RQ4513, was treated with 2 courses of BG/TMZ at 12 and 15 weeks after transplant. The ANC nadirs were 780 cells/μl at day 16 after the first treat-ment and 210 cells/μl at day 14 after the second treatment. The percentage of GFP+ and YFP+ granulocytes (Figure 3A) and monocytes (Figure 3C) increased from 4% to 40% (YFP) and 4% to 50% (GFP), but gradually declined back to 4% (GFP) and 10% (YFP) over a period of 50 days after the second dose of BG/TMZ. The absolute number of YFP-transduced granulocytes increased from 40 cells/μl before drug treatment to approximately 200 cells/μl after 2 courses of BG/TMZ, which suggests that YFP-transduced CD34+ cells were protected and expanded after BG/TMZ treatment (Figure 3B). Similarly, the percentage of GFP+ and YFP+ lymphocytes doubled after the second BG/TMZ administration, but GFP+ lymphocytes rapidly declined back to pretreatment levels (Figure 3D).

Figure 1Vector constructs. (A) For the GFP-MGMT* SIV-based lentiviral vectors, expression of a GFP-MGMT* fusion gene was driven by the promoter activity of the MSCV LTR. (B and C) For MGMT*-P2A-GFP (B) and MGMT*-P2A-YFP (C) SIV vectors (referred to herein as MGMT*-GFP and MGMT*-YFP, respectively), P2A was used to generate a multicistronic expression cassette inserted in the pCL20cSLFR MSCV-GFP backbone. The primer pairs (M3, M5, Y3, Y5, G3, and G5) used for construction of the vectors are described in Methods. (D) For MGMT*-P2A-HOXB4-T2A-GFP SIV vectors (referred to herein as MGMT*-HOXB4-GFP), T2A was used to generate a multicistronic expression cassette inserted in the pCL20cSLFR MSCV-GFP backbone. MGMT*, HOXB4, and GFP were amplified and 2A-tagged from their respective cDNAs using primer pairs M5, M3, H5, H3, G5, and G3 as described in Methods.

research article


The second animal, RQ4099, was treated with a single course of BG/TMZ at week 10 after transplant. The ANC nadir was 190 cells/μl at day 14 after treatment. The percentage of GFP+ and YFP+ granulocytes and monocytes increased from a steady-state level of 3%–5% to a peak of 25%–30%, but returned to baseline gene marking levels within 50 days after BG/TMZ administration (Figure 4A). In vivo enrichment of GFP+ and YFP+ B cells was also transient (Figure 4B). In T cells, GFP and YFP marking levels were 1% at steady state and peaked at 13% (GFP) and 23% (YFP) after BG/TMZ treatment. The proportion of gene marked T cells subsequently declined and stabilized at 4% (GFP) and 5% (YFP) (Figure 4B). The animal died approximately 3 months after BG/TMZ treatment due to pulmo-nary fibrosis. This complication is occasionally seen in animals not treated with BG/TMZ and is likely secondary to radiation received as a conditioning regimen prior to transplant.

The third animal, RQ4152, had pretreatment gene marking lev-els of 5% in granulocytes, 3% in monocytes, 10% in B cells, and 1% in T cells. The animal was treated with 7 courses of BG/TMZ, with TMZ doses ranging from 170 to 450 mg/m2/d. The ANC nadir was 850 cells/μl at day 6 after the first treatment (170 mg/m2 TMZ) and 7 cells/μl at day 9 after the last treatment (450 mg/m2 TMZ). After each course of drug treatment, there was a rapid increase in GFP marking in all lineages. The highest marking achieved after the last course of BG/TMZ treatment was 94% for myeloid cells and 57% for lymphoid lineages (Figure 5A). The proportion of

GFP+ cells subsequently declined and leveled off at 10% for granu-locytes, 5% for monocytes, 17% for B cells, and 30% for T cells at 150 days after the last treatment (Figure 5A). Correspondingly, the absolute number of transduced granulocytes increased after each drug treatment with, for example, levels of 80 cells/μl before the fourth course peaking to 698 cells/μl after the seventh course and stabilizing at approximately 250 cells/μl (Figure 5B). These results indicate that MGMT*-transduced cells were protected and expanded following multiple courses of BG/TMZ treatments with TMZ dose escalation up to 450 mg/m2. In pediatric patients, the maximum tolerated dose of TMZ without BG is 185 mg/m2 (29); therefore, gene therapy resulted in substantial dose escalation due to protection from hematopoietic toxicity. As shown by pharma-cokinetic data (Table 2), the absorption and clearance of TMZ in animal RQ4152 at the time of the seventh dose of TMZ was com-parable to that seen in treated pediatric patients. Moreover, the AUC for TMZ was about 2-fold greater than that seen in pediatric patients (Table 2), confirming that BM protection was achieved against BG/TMZ by MGMT* overexpression in this animal. This is particularly interesting given that BG is known to significantly enhance TMZ toxicity, necessitating TMZ dose reductions (30). Given the relative lack of nonhematopoietic toxicity seen in our series with BG/TMZ, we conclude that hematopoietic protection via transfer of MGMT* allows for substantial dose escalation and administration of repeated drug courses.

Figure 2In vivo selection in SIV-transduced animals RQ4876 and RQ4795 after multiple cycles of BG/BCNU and/or BG/TMZ. Both animals were trans-planted with CD34+ cells transduced with lentiviral GFP-MGMT* vector. (A) ANC, hemoglobin (HGB), and platelet (PLT) counts in PB over time for RQ4876 and RQ4795. (B) Percent GFP+ granulocytes (Gran), lymphocytes (Lymph), and monocytes (Mono) from PB for RQ4876 and RQ4795. Arrows denote cycles of drug administration; “†” indicates time of animal death.

research article


Polyclonal hematopoiesis is sustained after BG/TMZ treatment. The clonal composition before and after BG/TMZ treatment was analyzed by linear amplification–mediated PCR (LAM-PCR) in animals RQ4152 and RQ4513, which showed the highest in vivo gene marking levels. DNA isolated from PBLs before and after 1, 2, 6, and 7 courses of BG/TMZ was used for LAM-PCR. Gel electrophoresis showed contribution of multiple clones at all time points analyzed (Supplemental Figure 2). Individual bands from LAM-PCR products were picked from the gel and sequenced directly after PCR amplification. Using previously described selec-tion criteria (31), we retrieved and analyzed 254 independent unequivocal retroviral integration sites (ISs; 152 for RQ4152, Supplemental Table 1; and 102 for RQ4513, Supplemental Table 2). The majority of these ISs occurred within transcription units, mainly within introns (Table 3), consistent with data from pre-vious reports (31–33). Also, in agreement with previously pub-lished data (31–33), SIV integration events did not favor locations upstream or downstream of transcription units (Table 3), indicat-ing no preference for integration near transcription start sites. In animals RQ4152 and RQ4513, 7.9% and 3.9% of ISs identified, respectively, were in or near cancer-associated genes (CAGs; Table 3 and Supplemental Tables 1 and 2). We identified 8 clones that appeared repetitively before and after drug treatment, which sug-gests that the same clones contributed to hematopoiesis after in vivo selection with BG/TMZ.

In vivo treatment with BG/BCNU does not result in long-term chemo-protection and causes high levels of nonhematopoietic toxicity. In ani-mals RQ4876, RQ4513, and RQ4152, which showed evidence of increased gene marking after multiple BG/TMZ treatments, we tested whether chemoprotection could be achieved using BCNU as an alternative alkylating agent. Several months after the last dose of BG/TMZ, we administered 1–3 successive infusions of 120 mg/m2 BG and 20 mg/m2 BCNU (hereafter referred to as BG/BCNU). In animal RQ4876, a moderate but steady increase in marking levels was seen after each BG/BCNU dose. Pretreatment levels of 2% in granulocytes, 6% in lymphocytes, and 5% in monocytes peaked at 9%, 17%, and 10%, respectively, after the third dose (Figure 2B). BG/BCNU treatment also resulted in a rapid increase in GFP/YFP marking in all hematopoietic lineages in animal RQ4513, with maximal levels of 90% in granulocytes and monocytes, 60% in B cells, and 40% in T cells (Figure 3). A more modest effect was seen in animal RQ4152 (Figure 5).

All animals that received BG/BCNU eventually developed sub-stantial nonhematopoietic toxicity, in contrast to what was seen using BG/TMZ. RQ4876 developed severe tachypnea, fevers, and failure to thrive with weight loss 1 month after the third dose of BG/BCNU. X-rays revealed diffuse pulmonary infil-trates consistent with pneumonitis (Supplemental Figure 3A). Euthanasia was performed 35 days after the last administration of BG/BCNU. Animal RQ4513 developed severe pulmonary and

Figure 3In vivo selection in SIV-transduced animal RQ4513 after 2 cycles of BG/TMZ and 2 cycles of BG/BCNU. Rhesus competitive repopulation experi-ments were set up using equal numbers of mobilized PB CD34+ cells transduced with SIV vectors expressing MGMT*-YFP or MGMT*-HOXB4-GFP. (A) Percent GFP+ and YFP+ granulocytes from PB. (B) Absolute number of YFP+ granulocytes. (C) Percent GFP+ and YFP+ monocytes from PB. (D) Percent GFP+ and YFP+ B cells and T cells from PB. Arrows denote cycles of drug administration; “†” indicates time of animal death.

research article


gastrointestinal complications, and euthanasia was performed 2 weeks after the second cycle of BG/BCNU. Specimens collected at necropsy revealed extensive pulmonary edema and widespread colonic necrosis with pseudomembrane formation (Supplemen-tal Figure 3B). Animal RQ4152 was clinically well after receiv-ing 7 courses of BG/TMZ, but developed severe gastrointestinal complications and died 7 days after 1 course of BG/BCNU. At necropsy, specimens were consistent with a diagnosis of severe necrohemorrhagic colitis (Supplemental Figure 3C). While lower doses of BCNU may have been less toxic, we observed no signifi-cant decrease in complete blood count and no change in vector marking levels when a single test dose of 5 mg/m2 was adminis-tered in animal RQ4876 (data not shown).

Because all animals were euthanized shortly after BG/BCNU administration, long-term persistence of improved marking lev-els could not be assessed. In the BCNU-treated animal RQ4876, follow-up was possible for a slightly longer period of time after BCNU administration (about 80 days); in this animal, a clear trend down in marking levels was evident at the time of euthanasia (Fig-ure 3B), which suggests a lack of persistent chemoprotection with BCNU, as was seen after TMZ administration.

Absence of long-term chemoprotection in vivo is not caused by lack of transduction of long-term repopulating HSCs. We investigated wheth-er the lack of stable selection in vivo could be the result of poor transduction of long-term repopulating HSCs. Several lines of evi-dence indicated transduction of HSCs. First, in all animals, persis-tence of substantial expression of the GFP transgene in T cells, B cells, granulocytes, and monocytes for several months after trans-plant was an indication of HSC transduction (Figure 2B and Fig-ures 3–5). Second, in animal RQ4876, 10% of CD34+ cells isolated

from the BM 630 days after transplant were GFP+, as assessed by flow cytometry (data not shown). Third, we also obtained molecu-lar evidence of transduction of HSCs in animal RQ4876. Using DNA isolated from the BM of this animal more than 1 year after transplant, we performed LAM-PCR and isolated 10 ISs. Demon-stration of common lentiviral ISs in sorted subpopulations of the PB (T cells, B cells, and granulocytes) collected before BG/TMZ treatment (day 60 after transplant), after the last dose of BG/TMZ (day 374), and after the last dose of BG/BCNU (day 630) was per-formed by PCR using 1 vector long-terminal repeat (LTR) primer and 1 genomic primer spanning a region of each IS. We were able to detect 2 common ISs (Figure 6), providing a clear demonstra-tion of the transduction of HSCs contributing in the long term to the different hematopoietic lineages. Interestingly, of the 2 com-mon ISs, the one designated IS 1 (Figure 6) was undetectable prior to drug treatment, but its contribution to all lineages increased over time after BG/TMZ and BG/BCNU treatment, a characteris-tic of HSC dynamics. Overall, these observations rule out the lack of transduction of HSCs as a possible mechanism for the absence of long-term chemoprotection. Given the consistent data among the 5 animals described in the present study, we believe that these conclusions can be generalized to all animals.

HOXB4 does not confer HSC amplification in the rhesus macaque trans-plant model. HOXB4 is a potent HSC amplifier in murine transplan-tation models. In our competitive rhesus transplantation model, we investigated whether HOXB4 could improve engraftment of transduced HSC and progenitor cells in animals RQ4513, RQ4099, and RQ4152 (Table 1). Gene transfer efficiencies in CD34+ cells were similar between MGMT* and MGMT*-HOXB4-GFP vectors before cell infusion based on flow and CFU analysis. For animals RQ4513 and RQ4099, the GFP/YFP ratio allowed direct measure-ment of competitive engraftment of MGMT*-HOXB4-GFP– ver-sus MGMT*-transduced cells. In both animals, we observed a similar engraftment pattern. The initial engraftment levels for both MGMT*- and MGMT*-HOXB4-GFP–transduced granu-locytes at 2 weeks after transplant were similar in RQ4513 (17% versus 13%; Supplemental Figure 4A) and RQ4099 (23% versus 25%; Supplemental Figure 4B). Marking levels gradually declined for both compartments before they stabilized at a similar level. In RQ4513, MGMT*-HOXB4-GFP marking stabilized at 2.6%, and MGMT* marking stabilized at 2.9%, in granulocytes 78 days after transplant (Supplemental Figure 4A). Marking levels in granulo-cytes in RQ4099 stabilized at 2.5% for MGMT*-HOXB4-GFP and 3.3% for MGMT* at 66 days after transplant (Supplemental Fig-ure 4B). For animal RQ4152, because both MGMT* and MGMT*-HOXB4-GFP–transduced cells used GFP as an expression marker, Southern blot was used to distinguish the 2 compartments. Again, there was no difference in engraftment levels between MGMT*- and MGMT*-HOXB4-GFP–transduced cells (Supplemental Fig-ure 4C, lane 1). Moreover, we did not observe any advantage for in vivo selection in the cells containing the MGMT*-HOXB4-GFP

Figure 4In vivo selection in SIV-transduced animal RQ4099 after 1 cycle of BG/TMZ. Rhesus competitive repopulation experiments were set up using equal numbers of mobilized PB CD34+ cells transduced with SIV vectors expressing MGMT*-YFP or MGMT*-HOXB4-GFP. (A) Percent GFP+ and YFP+ granulocytes and monocytes from PB. (B) Percent GFP+ and YFP+ B cells and T cells from PB. Arrows denote cycles of drug administration.

research article


vector. For instance, in RQ4513 (Figure 3) and RQ4099 (Figure 4), marking levels in all lineages increased in parallel between MGMT*- and MGMT*-HOXB4-GFP–transduced cells after drug treatment. For RQ4152, there was also no difference in marking levels between MGMT*- and MGMT*-HOXB4-GFP–transduced cells after BG/TMZ treatment, as determined by Southern blot (Supplemental Figure 4C, lanes 2–5). At 8 months after trans-plant, HOXB4 expression was detected in MGMT*-HOXB4-GFP–marked PB cells sorted by flow cytometry (Supplemental Figure 5), which confirmed that the lack of amplification was not caused by absent expression of HOXB4.

DiscussionMGMT*-mediated in vivo selection with lentiviral vectors occurs mainly in progenitor cells, not HSCs. In the current study, we evaluated BG/TMZ and BG/BCNU for in vivo selection of CD34+ cells modi-fied with MGMT* lentiviral vectors to increase transduced blood cells in the clinically relevant rhesus macaque model. We observed that 2 animals with pretreatment steady-state in vivo marking lev-els of 2% and 0.2% showed transient increase in GFP expression in both myeloid and lymphoid lineages after 4 cycles of BG/TMZ treatment, indicating in vivo MGMT* gene expression and chemo-protection of progenitor cells with transient repopulating capac-ity. Additionally, 3 other animals with higher baseline marking levels (5% in PB granulocytes) showed a more persistent, yet still transient, increase in gene-marked cells in vivo in lymphoid and/or myeloid lineages after BG/TMZ administration. Additional cycles of chemotherapy using BCNU as an alternative alkylating agent resulted in similar increases in gene marking levels.

Overall, these results indicate that the relatively high transduction efficiencies achieved in CD34+ cells using lentiviral vectors support chemoprotection of progenitor cells, but not long-term repopulat-ing stem cells. Four mechanisms can be considered in account-ing for the lack of stable selection at the stem cell level. First, we demonstrated that the absence of persistent chemoprotection did not relate to the lack of transduction of long-term repopulating HSCs. Detection of transgene expression in the long term (close to 2 years in 1 animal) after transplant in both lymphoid and myeloid lineages was an indication of transduction of long-term repopulat-ing HSCs. In a prior study, 3 rhesus macaques — transplanted with CD34+ cells transduced using SIV vectors and conditions similar to those used in the present study — demonstrated significant vector marking and expression of the transgene for up to 4 years after transplant (34). In animal RQ4876 in the present study, 10% of CD34+ cells isolated from the BM at day 630 after transplant expressed the GFP transgene. More importantly, we demonstrat-ed 2 common lentiviral ISs in sorted subpopulations of the PB (T cells, B cells, and granulocytes) of RQ4876 collected up to 630 days after transplant. Of the 2 common ISs (Figure 6), 1 had char-acteristics of HSC, not progenitor cell, dynamics with amplifica-tion over time after BG/TMZ and BG/BCNU treatment. Overall, these observations indicate transduction of HSCs contributing in the long term to the different hematopoietic lineages.

Second, in spite of the observed lack of selection with TMZ, it is possible that selection of HSCs could have been achieved with BCNU at doses lower than 20 mg/m2, as toxicity prevented long-term follow-up. In animal RQ4876, we administered a lower dose of BCNU (5 mg/m2), but no significant decrease in complete blood count and no change in vector marking levels was observed. While titrating the dose of BCNU may be possible in rhesus macaques for selection of transduced HSCs with decreased toxicity, it is unlikely that this drug could be used safely in patients for which individu-alized dose titration would not be possible.

Third, we considered the possibility that the lack of HSC selec-tion could be related to the deleterious effect of high murine stem cell virus–driven (MSCV-driven) expression of MGMT* on cellular proliferation and engraftment, or, alternatively, to inad-equate MGMT expression from this promoter in HSCs. As recently reported by Milsom and colleagues, while MGMT* expression by weaker cellular promoters (e.g., phosphoglycerate kinase) was suf-ficient for in vivo selection, viral promoter–driven high expression of MGMT* led to increased localization to the nucleus/chroma-tin, resulting in competitive repopulation defects in HSCs (35). However, the toxicity was observed with the SFFV viral promoter, causing a 10-fold higher level of MGMT* than with the MSCV promoter used in the present study. In addition, in our present study, MGMT* activity was sufficient to allow transient chemo-protection, but the low BG-noninhibitable activity detected in our transplanted animals (e.g., 7 fmol [3H]-methylated protein/mg protein detected in RQ4876) suggested that the potential toxicity

Figure 5In vivo selection in SIV-transduced animal RQ4152 after multiple cycles of BG/TMZ and 1 cycle of BG/BCNU. Rhesus competitive repopula-tion experiments were set up using equal numbers of mobilized PB CD34+ cells transduced with SIV vectors expressing MGMT*-GFP or MGMT*-HOXB4-GFP. (A) Percent GFP+ granulocytes, B cells, T cells, and monocytes from PB for RQ4152. (B) Absolute number of GFP+ granulocytes. Arrows denote cycles of drug administration.

research article


of high MGMT* expression from the lentiviral constructs we used is an unlikely explanation for the lack of stable enrichment of gene modified cells in rhesus macaques.

In contrast, the internal MSCV promoter used in the present study, although active in the rhesus CD34+ progenitor cells and blood progeny, may have not provided adequate expression of the MGMT* transgene in HSCs. The MSCV promoter was chosen at the outset of these experiments based on our extensive experience with it in the murine system, where we have previously shown it to be active in HSCs (36–38). Interestingly, a recent study showed that while the MSCV LTR is expressed in undifferentiated mouse ES cells, it has much lower transcriptional activity than that of the human elongation factor–1α (EF1α) promoter or the hybrid CAG promoter (CMV enhancer, chicken β-actin promoter and intron, and globin polyadenylation signal; ref. 39). In the context of self-inactivating (SIN) HIV-based lentiviral vectors, the EF1α promoter gave about a 6-fold increase in expression relative to that of the MSCV LTR, as judged by GFP fluorescence, while the CAG vector gave 2- to 3-fold better expression (39). The promoter MND may also offer advantages for expression in HSCs. It includes deletion of the negative control region in the 5′ region of the U3 portion of the LTR, and it contains 2 copies of the 72-bp enhancer repeat, com-pared with 1 copy in MSCV (40–43). Another promoter with poten-tial activity in HSCs is the ubiquitin-C promoter. SIN lentiviral vec-tors using this promoter to drive GFP expression were previously found to express well in the primitive embryonic blastula cells of mice after injection of vector into single-cell embryos (44). Togeth-er, these data raise the possibility that either of the latter regulatory elements may provide better expression than the MSCV promoter in HSCs, thus allowing better long-term in vivo selection.

Fourth, the lack of selection at the stem cell level may be attrib-uted to the ability of untransduced stem cells to resist the cytotox-ic effects of TMZ. The cytotoxicity of TMZ is largely attributed to the induction of single-strand methyl adducts at the O6 position of guanines, leading to mispairing with thymidine residues. When TMZ-treated cells divide, DNA replication is inhibited, trigger-ing apoptosis and cell death. In contrast to rapidly dividing pro-genitor cells, HSCs cycle slowly in vivo in large animals (once per 23–36 weeks at steady state; ref. 45); therefore, the TMZ-induced O6-methylguanine DNA adducts may persist longer in these cells with no deleterious effects. During this period of quiescence, in the absence of BG, endogenous MGMT levels would be expected

to recover, leading to repair of O6-methylguanine DNA adducts in both transduced and untransduced HSCs. When the untrans-duced HSCs eventually divide, the TMZ-induced DNA lesions may have been repaired, preventing the cells from undergoing apop-tosis. Given the marked differences in estimates of the average HSC replication rate between mice (once per 2.5 weeks; refs. 45, 46) and large animals (once per 23–36 weeks; ref. 45), the same reasoning could also explain why in vivo selection of HSCs using MGMT* was successfully achieved in mice (10–12). In our rhesus macaque study, BG was given once daily for 5 days; perhaps con-tinuous administration of BG for a longer period of time could have prevented recovery of endogenous MGMT in slowly dividing HSCs and allowed for a stable increase in gene marking levels in vivo. Another option would be to design a treatment protocol that forces HSC cycling using appropriate hematopoietic cytokines during the period of cytotoxic drug treatment.

An alternative explanation for the increased resistance of untrans-duced HSCs to BG inactivation may be related to higher levels of endogenous MGMT compared with more mature progenitor cells. Indeed, high levels of expression of molecules such as ABCG2, MDR-1, and ALDH in HSCs (47–50) are postulated to be of impor-tance in protecting HSCs from genotoxic or metabolic damage. Moreover, in highly purified murine long-term repopulating HSCs (defined as CD34–Lin–c-Kit+Sca1+), MGMT expression was 2–3 times higher than that present in more mature progenitor cells (defined as Lin–c-Kit+Sca1–), as determined previously by query of a GEO DataSet Browser for microarray data of MGMT expression in hematopoietic cells (51). Optimization of inhibition of endogenous HSC MGMT through a pharmacokinetic and pharmacodynamic approach could improve in vivo selection in HSCs.

BG/BCNU has more nonhematopoietic toxicity than does BG/TMZ for MGMT*-mediated in vivo selection. An important consideration regard-ing the ultimate use of the MGMT* selection system in humans is whether it can be used with acceptable toxicity. In our rhesus macaque study, euthanasia was performed in 3 of 3 (100%) animals that received BG/BCNU. In contrast, animal RQ4795 survived treatment with BG/TMZ alone without complications, and animal RQ4099, also treated with BG/TMZ alone, was euthanized follow-ing complications attributed to radiation-induced pulmonary fibro-sis. These observations indicate a higher toxicity of the alkylating agent BCNU compared with TMZ. Gastrointestinal complications were observed with BCNU, including necrohemorrhagic colitis in animal RQ4152 and colonic necrosis with pseudomembrane forma-tion in animal RQ4513. Pulmonary complications were also seen in 2 animals, including pulmonary congestion/edema (RQ4513) and infections (RQ4876). These complications have never or only rarely been detected in more than 200 transplanted rhesus macaques at

Table 2Pharmacokinetic study of RQ4152 after the seventh course of TMZ/BG treatment

Parameter RQ4152 Pediatric populationA

TMZ clearance (l/h/m2) 5.4 4.9 (1.6–10.8)Cmax (mg/l) 24.6 9.1 (4.3–19.0)Tmax (h) 1.0 1.2 (0.2–2.9)Ka (h–1) 1.6 2.4 (0.1–31.6)TMZ AUC value (μg/ml•h) 65.1 37.7 (18.6–120.3)MTIC AUC value (μg/ml•h) 1.9 1.0 (0.1–2.1)

Cmax, maximum concentration; Tmax, maximum time; Ka, absorption rate constant; MTIC, 3-methyl-(triazen-1-yl)imidazole-4-car-boxamide, an active species to which TMZ is spontaneously hydrolyzed at physiologic pH. AMean (range) in a pediatric population receiving approximately 185 mg/m2 TMZ (29).

Table 3IS distribution and association with CAGs for RQ4152 and RQ4513

Parameter RQ4152 RQ4513Transcription units 70% 77%Introns 64% 71%Exons 6% 6%Within 10 kb upstream of gene 4.6% 3.9%Within 10 kb downstream of gene 2.6% 3.9%In or near CAGs 7.9% 3.9%

research article


our institutions, which suggests that they are causally related to the administration of BG/BCNU. In contrast, TMZ toxicity was highly specific for hematopoietic cells, and multiple courses were well tol-erated without cumulative or nonhematopoietic toxicity.

HOXB4 does not provide a proliferative advantage to transduced long-term repopulating cells. In addition to drug resistance genes for selection of transduced cells after transplantation, retroviral vec-tors armed with genes that give a proliferative advantage to the transduced cells have also been proposed to improve in vivo gene marking levels (28). Genes that favor self renewal of HSCs have been prime candidates to promote selective growth advantage of transduced HSCs for improved gene therapy. One particular tran-scription factor, HOXB4, has been extensively studied as a way of increasing the self renewal of HSCs and hematopoietic progenitor cells in mice (24, 25, 52), humans, and large animal models (27) both in vivo and in vitro.

In our primate competitive repopulation assay, we transduced monkey CD34+ cells with either a MGMT*-HOXB4-GFP vector or a vector expressing only the MGMT* gene and analyzed the com-petitive repopulating ability of these cells in vivo. We found no substantial difference in engraftment levels between MGMT*- and MGMT*-HOXB4-GFP–transduced cells both before and after che-motherapy treatment up to 400 days after transplant. Our data are consistent with those of Zhang, Kiem, and colleagues, who found that, in contrast to mouse studies, no significant advantage for the HOXB4-transduced cells was observed in the long term after transplant in a similar nonhuman primate competitive study (27). The same group also recently reported what we believe to be the first instances of leukemia linked to HOXB4 expression, both in the original group of monkeys now followed longer term and in dogs that received cells transduced with a HOXB4-expressing ret-roviral vector (53). Some 2 years after transplant, 3 of 4 animals (2 of 2 dogs and 1 of 2 nonhuman primates) developed acute myeloid leukemias, and several lines of evidence have strongly implicated HOXB4 in leukemogenesis (53). In animal RQ4152, euthanized as a result of complications of BG/BCNU treatment, we demonstrat-ed by direct sequencing no occurrence of monoclonality, at least over the period of observation of approximately 12 months.

MGMT* lentiviral vectors can be used for TMZ dose intensification. TMZ is a chemotherapeutic alkylating agent with demonstrated

efficacy for the treatment of malignant gliomas. A recent study by Stupp et al. revealed that TMZ in combination with radiation therapy yielded a statistically significantly greater prolongation of the survival time compared with radiation alone in patients with newly diagnosed gliomas (23). Increased doses of TMZ may lead to further increases in survival, but dose intensification is precluded because of increased toxicity. In the present study, we showed that BM protection was achieved by MGMT* overexpres-sion after multiple courses of BG/TMZ treatments with TMZ dose escalation up to 450 mg/m2, twice the maxi-mum tolerated dose of TMZ in pediatric patients. This is potentially significant because the use of BG in patients to diminish MGMT activity and drug resistance in tumor cells has been associated with increased hematopoietic toxicity and has necessitated TMZ dose reductions (30). Our findings indicate that drug resistance in 5%–10% of myeloid progenitors allows for increased hematopoietic recovery and increased tolerance to TMZ. The transient chemoprotection observed in this study may be useful

for patients with malignant brain tumors treated with both BG and TMZ dose intensification in order to overcome tumor cell drug resistance. While cumulative genotoxicity resulting from the combined use of alkylating agents (e.g., TMZ) and lentiviral vec-tors needs to considered, our study revealed no evidence of clonal dominance or leukemic transformation using this approach, at least over the period of observation covered.

MethodsCollection of PB HSCs from rhesus macaques. All 5 rhesus macaques (Macaca mulatta) used in these studies were housed and handled in accordance with the guidelines set by the Committee on Care and Use of Laboratory Animals of the Institute of Laboratory Animal Resources, National Research Coun-cil (DHHS publication no. NIH 85-23). Animals RQ4876 and RQ4795 were treated under a protocol approved by the Animal Care and Use Committee of the NHLBI. Animals RQ4513, RQ4099, and RQ4152 were housed in the Animal Resource Center of St. Jude Children’s Research Hospital and were treated under a protocol approved by the St. Jude IACUC.

For stem cell mobilization, animals RQ4876 and RQ4795 were mobilized using a single subcutaneous dose of 1 mg/kg AMD3100 (Genzyme), and PB cells were collected by leukapheresis 3 hours after AMD3100 administra-tion as described previously (54). The other 3 animals, RQ4513, RQ4099, and RQ4152, were given G-CSF at 10 μg/kg/d and SCF at 50 μg/kg/d (Amgen) through subcutaneous injection for 3 days and twice daily on day 4. For RQ4513 and RQ4099, mobilized PB cells were collected by leuka-pheresis on day 5 as described previously (55); for RQ4152, BM was aspi-rated from the posterior iliac crest. In animal RQ4152, BM was harvested after in vivo cytokine administration based on our previous observation that in vivo manipulation of HSCs prior to BM harvest can improve ret-roviral gene transfer into HSCs by promoting cell cycling and increasing receptor density in nonhuman primate models (56).

For each cell fraction (AMD3100-mobilized PB cells, G-CSF and SCF–mobilized PB cells, and BM), PBMCs were isolated using density gradient centrifugation over lymphocyte separation media (LSM; ICN Biomedicals/Cappel). CD34+ cells were enriched using the 12.8 IgM anti-CD34 biotinyl-ated antibody and MACS streptavidin microbeads (Miltenyi Biotec) per the manufacturer’s instructions. Alternatively, CD34+ cells were purified from BM or PB leukapheresis product using a PE-conjugated CD34 antibody (clone 563; BD Biosciences — Pharmingen) on a Miltenyi CliniMacs Column (Miltenyi). The number of CD34+ cells isolated represented approximately

Figure 6Transduction of long-term repopulating rhesus macaque HSCs with lentiviral vectors. Using DNA isolated from the BM of rhesus macaque RQ4876 more than 1 year after transplant, we performed LAM-PCR and isolated 10 ISs. Dem-onstration of common lentiviral ISs in sorted subpopulations of the PB — T cells, B cells, and granulocytes — collected before and after BG/TMZ or BG/BCNU selection was subsequently performed by PCR using 1 vector LTR primer and 1 genomic primer spanning a region of each IS. We detected 2 common ISs, providing a clear demonstration of the transduction of HSCs contributing in the long term to the different hematopoietic lineages. DNA extracted from wbcs of a nontransplanted animal was used as a negative control. Each DNA sample was amplified using β-actin–specific primers as a positive control.

research article


1% of the mononuclear cells derived from the nonhuman primate BM or PB after mobilization. Purities of 90%–98% CD34+ cells were obtained.

Vector construction. For GFP-MGMT* SIV vectors (Figure 1A), the open reading frame of the fusion protein GFP-MGMT* was used to replace the GFP cDNA sequence in the parental SIVmac1A11 vector plasmid pCL20cSLFR MSCV-GFP constructed by Hanawa et al. (57) using EcoRI and NotI restriction enzymes. Construction of the fusion protein was described previously (58).

For MGMT*-GFP SIV vectors (Figure 1B), P2A was used to generate a multicistronic expression cassette inserted in the pCL20cSLFR MSCV-GFP backbone. In contrast to multicistronic vectors using internal ribo-some entry sites, where multiple proteins are often not expressed at the same level, multicistronic retroviral vectors using 2A peptides to link 2 or several genes have previously been shown to result in stoichiomet-ric production of each protein (59). MGMT* and GFP were amplified and P2A tagged from their cDNAs using primer pairs M5 (5′-TCTC-TAGGCGCCGGAATTCCGGATGGACAAGGATTGTGAAATGAAAC-GCACCACACTGGACAGCCCTTTGGGGAAGCTGGAG-3 ′), M3 (5′- GGGACCGGGGTTTTCTTCCACGTCTCCTGCTTGCTTTAACAGAGAGAAGTTC-GTGGCTCCGGAGTTTCGGCCAGCAGGCGGGGAGCCCGA-3′), G5 (5′-TCCGGAGAGGGCAGAGGAAGTCTGCTAACATGCGGTGACGTCGAG-GAGAATCCTGGCCCAATGGTGAGCAAGGGCGAGGAGCTGTTC-3′), and G3 (5′-CCATCGATGGTTACTTGTACAGCTCGTCCATGCCGAGA-GTGATCCCGGCGGCGGTCACGAACTCC-3′). A secondary PCR reaction was performed using MGMT*-P2A and P2A-GFP as templates and M5 and G3 as primers. MGMT*-YFP SIV vectors (Figure 1C) were constructed using a similar strategy with primer pairs M5, M3, Y5 (5′-TCCGGAGCCAC-GAACTTCTCTCTGTTAAAGCAAGCAGGAGACGTGGAAGAAAACCCC-GGTCCCATGGTGAGCAAGGGCGAGGAGCTGTTC-3′), and Y3 (5′-CCATCGATGGTTACTTGTACAGCTCGTCCATGCCGAGAGT-GATCCCGGCGGCGGTCACGAACTCC-3′).

For MGMT*-HOXB4-GFP SIV vectors (Figure 1D), T2A was used to generate a multicistronic expression cassette inserted in the pCL20cSLFR MSCV-GFP backbone. MGMT*, HOXB4, and GFP were amplified and 2A tagged from their respective cDNAs using primer pairs M5, M3, H5 (5′-TCCGGAGCCACGAACTTCTCTCTGTTAAAGCAAGCAGGAGAC-GTGGAAGAAAACCCCGGTCCCATGGCTATGAGTTCTTTTTTGAT-CAACTCA-3′), H3 (5′-TGGGCCAGGATTCTCCTCGACGTCACCGCAT-GTTAGCAGACTTCCTCTGCCCTCTCCGGAGAGCGCGCGGGGGCCTCCATT-3′), G5, and G3. A secondary PCR reaction was performed using MGMT*-P2A and P2A-HOXB4 as templates and M5 and H3 as primers to gen-erate the product MGMT*-P2A-HOXB4-T2A, which was used as the template together with T2A-GFP for the tertiary PCR using primers M5 and G3 to generate the expression cassette. The expression cassettes were then digested with EcoRI and ClaI and subcloned into the pCL20cSLFR MSCV-GFP backbone.

Virus production. Amphotropic-pseudotyped SIV vector particle con-ditioned medium was produced as described previously (57). Briefly, 1 × 107 293T cells were seeded onto 10-cm dishes and transfected with a total of 44 μg DNA per dish containing the following plasmids: 12 μg pCAG-SIV gprre; 4 μg pCAG4-RTR-SIV; 4 μg pCAG-Ampho; and 24 μg pCL20cSLFR GFP-MGMT*, MGMT*-YFP, MGMT*-GFP, or MGMT*-HOXB4-GFP. The cells were washed with PBS 18 hours later and cul-tured in fresh medium for an additional 24 hours. Conditioned medium was then harvested, filtered through a 22-μm nylon filter (Millipore) to remove cellular debris before transduction, concentrated by ultracentri-fugation (9,000 g for 3 hours), resuspended in X-vivo medium (BioWhit-taker Inc.), and frozen at –80°C. Each vector preparation was tested for replication-competent lentivirus as described previously (57). In brief, 5 × 104 HeLa cells were transduced with an equal amount of SIV particles

and grown for 2–3 weeks. Conditioned medium was then tested for the presence of SIV p27 using an ELISA kit (Coulter).

Transduction of rhesus repopulating HSCs. For animals RQ4876 and RQ4795, CD34+ cells were cultured at a starting concentration of 1 × 106 cells/ml in filtered vector supernatant (MOI of 1) supplemented with 100 ng/ml recombinant human SCF (rhuSCF; Amgen), 100 ng/ml recom-binant human FMS-like tyrosine kinase-3 ligand (rhuFlt-3L; Immunex), 100 ng/ml thrombopoietin (TPO; R&D Systems), and 4 μg/ml protamine sulfate (Sigma-Aldrich) in 6-well plates previously coated with the CH-296 fragment of fibronectin (Retronectin; TaKaRa) per the manufacturers’ instructions. Every 24 hours, nonadherent cells were collected, spun down, resuspended in fresh vector supernatant and cytokines, and added back to the same fibronectin-coated plates. At the end of the 96-hour incubation period, transduced cells were removed from the plates using cell scrapers.

For animals RQ4513, RQ4099, and RQ4152, CD34+ cells were prestimulat-ed at 37°C in DMEM supplemented with 10% FCS (HyClone) and 100 ng/ml each of SCF, rhuFlt-3L, and TPO for 24 hours and then transduced on Retronectin-coated plates with vector-containing medium (MOI of 1) for 48 hours, with fresh medium containing cytokines and vector particles provided at the midpoint.

Cells were infused back into each animal conditioned with 2 doses of 450–500 cGy total body irradiation. After 24 hours, 5 μg/kg G-CSF was initiated i.v. daily until the total wbc count reached 6,000 cells/μl. Standard supportive care was administered, including blood product transfusions, fluid and electrolyte management, and antibiotics as needed. Hematopoi-etic recovery was monitored by daily complete blood counts.

Drug treatment. All drugs were freshly prepared on the day of drug treat-ment. BG (50 mg; Sigma-Aldrich) was dissolved in 10 ml polyethylene glycol (Fisher Scientific) and further diluted to 2.5 mg/ml with prewarmed (37°C) saline. BCNU (Bristol-Myers Squibb Co.) was diluted according to the manu-facturer’s instructions and was further diluted in normal saline to a total vol-ume of 18 ml before infusion. TMZ capsules (Schering-Plough) were opened, their content was weighed, and capsules were diluted in sterile H2O.

For drug treatment to evaluate in vitro hematopoietic chemoprotection, transduced CD34+ cells were exposed in vitro for 1 hour to BG (0–10 μM) followed by exposure to various concentrations of TMZ (0–200 μM) or BCNU (0–10 μM) for an additional 2 hours. Cells were then washed and cultured for an additional 10 days.

For in vivo drug treatment, animals were anesthetized, and Kytril (Roche) was administered prophylactically before drug administration. BG was infused i.v. into the animal at a dose of 120 mg/m2 over 20 minutes. TMZ was administered at a concentration of 245 mg/m2 for animals RQ4876 and RQ4795 and in 7 escalating doses for animals RQ4513, RQ4099 and RQ4152 (170, 202, 230, 284, 325, 365, and 450 mg/m2) via nasogastric tube or by direct oral administration of capsules over 5 minutes 1 hour after the end of BG infusion. Treatment was repeated daily for 5 consecutive days. BCNU was administered i.v. over a 5-minute period at a dose of 20 mg/m2 30 minutes after the end of BG infusion. A repeat dose of BG was adminis-tered i.v. at a concentration of 120 mg/m2 approximately 7 hours after the end of the first BG infusion.

Analysis of transduction efficiency. For RQ4876 and RQ4795, we used SIV vectors expressing GFP-MGMT* fusion protein, and transduction effi-ciency was determined by flow cytometry analysis using a Coulter FC500 analyzer (Beckman Coulter).

For RQ4513 and RQ4099, MGMT*-YFP versus MGMT*-HOXB4-GFP vectors were used, allowing for GFP versus YFP identification of cells in the competitive repopulating study. Flow cytometric analysis was performed for GFP and YFP expression using a FACSVantage SE instrument with FACSDiva software (version 5.0; BD). To quantitate GFP+ and YFP+ cells in immunophenotypically defined subsets of PB cells, the following PE-con-

research article


jugated antibodies were used: CD11b (clone ICRF44), CD14 (clone MSE2), CD16 (clone 3G8), CD3 (clone SP34-2), CD4 (clone MT310), CD8 (clone DK25), and CD20 (clone 2H7; all from BD Biosciences — Pharmingen).

For RQ4152, both vectors used GFP as expression marker. Therefore, Southern blot analysis was used to distinguish the 2 subsets of cells. Purified PB DNA was digested with BsrGI, which liberated a 2.3-kb frag-ment for the MGMT*-HOXB4-GFP vector and a 1.5-kb fragment for the MGMT*-GFP vector, and hybridized with a probe corresponding to a par-tial MGMT* sequence.

Evaluation of vector ISs. Purified PB DNA was prepared at different time points after transplant using Puregene DNA extraction kit (Gentra Sys-tems), and LAM-PCR was performed as described previously (57). Briefly, 100 ng DNA was amplified using a SIV-LTR–specific 5′-biotinylated prim-er and digested with ApoI enzyme and then ligated to a linker cassette. Nested PCR was performed using SIV-LTR–specific and linker-specific primers. Next, 5 μl of the PCR products were separated on a high-reso-lution spreadex gel (Elchrom Scientific), and each band was then picked from the gel using a bandpick (Elchrom Scientific) and amplified using the nested PCR primer pair and sequenced directly. We considered a sequence as a genuine retroviral IS only if it (a) juxtaposed to the vector LTR, (b) yielded a unique best hit by BLAT software (http://genome.ucsc.edu/cgi-bin/hgBlat), and (c) showed at least 90% identity to the July 2003 human genome assembly (60, 61). Annotations for CAGs were determined using the Sanger Institute Human Cancer Gene Census Database (http://www.sanger.ac.uk/genetics/CGP/Census) (62).

Demonstration of common vector ISs. Using DNA isolated from the BM of rhesus macaque RQ4876 more than 1 year after transplant, we performed LAM-PCR and isolated 10 ISs. Demonstration of common lentiviral ISs in sorted subpopulations of the PB (T cells, B cells, and granulocytes) collected before in vivo selection (2 months after transplant), after BG/TMZ selection (12 months), and after BG/BCNU treatment (31 months) was performed by PCR using 1 vector LTR primer (5′-AGAGCGACTGACTACATAGC-3′)

and 1 genomic primer spanning a region of each IS (IS 1, 5′-TTAGATGT-GCTGGGGTGGCATAAGAAGC-3′; IS 2, 5′-AATTCTGCAGATATCCAG-CACAGTGG-3′). Each DNA sample was amplified using β-actin–specific primers as positive control (forward, 5′-AAGAGAGGCATCCTCACCCT-3′; reverse, 5′-GAAGGTGTGGTGCCAGATTTTC-3′).

MGMT activity assay. Transduced or naive rhesus macaque BM cells were sonicated (30 seconds ×2 in 50 mM Tris-HCl, pH 7.6; 5 mM DTT; and 0.1 mM EDTA; XL-series sonicator; Misonix Inc.) and centrifuged to collect supernatant. MGMT activity was assayed by incubation of supernatants for 30 minutes at 37°C with [3H]-methylated DNA substrate in the presence or absence of 200 μM BG, followed by acid hydrolysis and separation of free bases on HPLC; activity was expressed as O6-methylguanine transferred to MGMT (fmol/mg protein; Bradford protein assay; ref. 63).

Statistics. Data were analyzed using the unpaired Student’s t test using Sigma Plot software. All statistical tests were 2 sided. A P value of 0.05 was considered to be statistically significant.

AcknowledgmentsWe thank Arthur Nienhuis for scientific discussion. This work was supported in part by the intramural research program of NHLBI and NIAID, NIH; by NHLBI grant P01 HL 53749; by Cancer Center Support Grant P30 CA 21765; by the Assisi Foundation of Mem-phis; and by the American Lebanese Syrian Associated Charities.

Received for publication September 19, 2008, and accepted in revised form April 15, 2009.

Address correspondence to: Andre Larochelle, National Heart, Lung and Blood Institute, National Institutes of Health, 9000 Rockville Pike, Building 10CRC, Room 3E-5256, Bethesda, Mary-land 20892, USA. Phone: (301) 451-7139; Fax: (301) 496-8396; E-mail: [email protected].

1. Nienhuis, A.W., Dunbar, C.E., and Sorrentino, B.P. 2006. Genotoxicity of retroviral integration in hematopoietic cells. Mol. Ther. 13:1031–1049.

2. Nienhuis, A.W. 2008. Development of gene therapy for blood disorders. Blood. 111:4431–4444.

3. Ferguson, C., Larochelle, A., and Dunbar, C.E. 2005. Hematopoietic stem cell gene therapy: dead or alive? Trends Biotechnol. 23:589–597.

4. Horn, P.A., Morris, J.C., Neff, T., and Kiem, H.P. 2004. Stem cell gene transfer--efficacy and safety in large animal studies. Mol. Ther. 10:417–431.

5. Cavazzana-Calvo, M., et al. 2000. Gene therapy of human severe combined immunodeficiency (SCID)-X1 disease. Science. 288:669–672.

6. Neff, T., Beard, B.C., and Kiem, H.P. 2006. Survival of the fittest: in vivo selection and stem cell gene therapy. Blood. 107:1751–1760.

7. Podda, S., et al. 1992. Transfer and expression of the human multiple drug resistance gene into live mice. Proc. Natl. Acad. Sci. U. S. A. 89:9676–9680.

8. Sorrentino, B.P., et al. 1992. Selection of drug resis-tant bone marrow cells in vivo after retroviral trans-fer of human MDR1. Science. 257:99–103.

9. Allay, J.A., et al. 1998. In vivo selection of retroviral-ly transduced hematopoietic stem cells. Nat. Med. 4:1136–1143.

10. Davis, B.M., Koc, O.N., and Gerson, S.L. 2000. Limiting numbers of G156A O(6)-methylguanine-DNA meth-yltransferase- transduced marrow progenitors repop-ulate nonmyeloablated mice after drug selection. Blood. 95:3078–3084.

11. Ragg, S., et al. 2000. Direct reversal of DNA damage by mutant methyltransferase protein protects mice against dose-intensified chemotherapy and leads to in vivo selection of hematopoietic stem cells. Cancer Res. 60:5187–5195.

12. Sawai, N., et al. 2001. Protection and in vivo selec-tion of hematopoietic stem cells using temozolo-mide, O(6)-benzylguanine, and an alkyltransferase-expressing retroviral vector. Mol. Ther. 3:78–87.

13. Persons, D.A., et al. 2004. Transient in vivo selec-tion of transduced peripheral blood cells using antifolate drug selection in rhesus macaques that received transplants with hematopoietic stem cells expressing dihydrofolate reductase vectors. Blood. 103:796–803.

14. Hibino, H., et al. 1999. The common marmoset as a target preclinical primate model for cytokine and gene therapy studies. Blood. 93:2839–2848.

15. Licht, T., et al. 2002. Drug selection with paclitaxel restores expression of linked IL-2 receptor gamma -chain and multidrug resistance (MDR1) transgenes in canine bone marrow. Proc. Natl. Acad. Sci. U. S. A. 99:3123–3128.

16. Crone, T.M., Kanugula, S., and Pegg, A.E. 1995. Mutations in the Ada O6-alkylguanine-DNA alkyl-transferase conferring sensitivity to inactivation by O6-benzylguanine and 2,4-diamino-6-benzyloxy-5-nitrosopyrimidine. Carcinogenesis. 16:1687–1692.

17. Loktionova, N.A., Xu-Welliver, M., Crone, T.M., Kanugula, S., and Pegg, A.E. 1999. Protection of CHO cells by mutant forms of O6-alkylguanine-DNA alkyltransferase from killing by 1,3-bis-(2-chloroethyl)-1-nitrosourea (BCNU) plus O6-ben-zylguanine or O6-benzyl-8-oxoguanine. Biochem. Pharmacol. 58:237–244.

18. Gerull, S., Beard, B.C., Peterson, L.J., Neff, T., and Kiem, H.P. 2007. In vivo selection and chemopro-tection after drug resistance gene therapy in a non-myeloablative allogeneic transplantation setting in dogs. Hum. Gene Ther. 18:451–456.

19. Neff, T., et al. 2003. Methylguanine methyltrans-

ferase-mediated in vivo selection and chemopro-tection of allogeneic stem cells in a large-animal model. J. Clin. Invest. 112:1581–1588.

20. Neff, T., et al. 2005. Polyclonal chemoprotection against temozolomide in a large-animal model of drug resistance gene therapy. Blood. 105:997–1002.

21. Pollok, K.E., et al. 2003. In vivo selection of human hematopoietic cells in a xenograft model using com-bined pharmacologic and genetic manipulations. Hum. Gene Ther. 14:1703–1714.

22. Zielske, S.P., Reese, J.S., Lingas, K.T., Donze, J.R., and Gerson, S.L. 2003. In vivo selection of MGMT(P140K) lentivirus-transduced human NOD/SCID repopulating cells without pretrans-plant irradiation conditioning. J. Clin. Invest. 112:1561–1570.

23. Stupp, R., et al. 2005. Radiotherapy plus concomi-tant and adjuvant temozolomide for glioblastoma. N. Engl. J. Med. 352:987–996.

24. Perkins, A.C., and Cory, S. 1993. Conditional immortalization of mouse myelomonocytic, mega-karyocytic and mast cell progenitors by the Hox-2.4 homeobox gene. EMBO J. 12:3835–3846.

25. Sauvageau, G., et al. 1995. Overexpression of HOXB4 in hematopoietic cells causes the selective expansion of more primitive populations in vitro and in vivo. Genes Dev. 9:1753–1765.

26. Antonchuk, J., Sauvageau, G., and Humphries, R.K. 2002. HOXB4-induced expansion of adult hemato-poietic stem cells ex vivo. Cell. 109:39–45.

27. Zhang, X.B., et al. 2006. Differential effects of HOXB4 on nonhuman primate short- and long-term repopulating cells. PLoS Med. 3:e173.

28. Sawai, N., Persons, D.A., Zhou, S., Lu, T., and Sor-rentino, B.P. 2003. Reduction in hematopoietic stem cell numbers with in vivo drug selection can

research article


be partially abrogated by HOXB4 gene expression. Mol. Ther. 8:376–384.

29. Panetta, J.C., et al. 2003. Population pharmacoki-netics of temozolomide and metabolites in infants and children with primary central nervous system tumors. Cancer Chemother. Pharmacol. 52:435–441.

30. Quinn, J.A., et al. 2005. Phase I trial of temozolo-mide plus O6-benzylguanine for patients with recurrent or progressive malignant glioma. J. Clin. Oncol. 23:7178–7187.

31. Hematti, P., et al. 2004. Distinct genomic integra-tion of MLV and SIV vectors in primate hematopoi-etic stem and progenitor cells. PLoS Biol. 2:e423.

32. Mitchell, R.S., et al. 2004. Retroviral DNA integra-tion: ASLV, HIV, and MLV show distinct target site preferences. PLoS Biol. 2:E234.

33. Bushman, F., et al. 2005. Genome-wide analysis of retroviral DNA integration. Nat. Rev. Microbiol. 3:848–858.

34. Kim, Y.J., et al. 2009. Sustained high level polyclonal hematopoietic marking and transgene expression four years following autologous transplantation of rhesus macaques with SIV lentiviral vector trans-duced CD34+ cells. Blood. Online publication ahead of print. doi:10.1182/blood-2008-10-185199.

35. Milsom, M.D., et al. 2008. Reciprocal relationship between O6-methylguanine-DNA methyltransfer-ase P140K expression level and chemoprotection of hematopoietic stem cells. Cancer Res. 68:6171–6180.

36. Hanawa, H., et al. 2004. Extended beta-globin locus control region elements promote consistent therapeutic expression of a gamma-globin len-tiviral vector in murine beta-thalassemia. Blood. 104:2281–2290.

37. Hanawa, H., Persons, D.A., and Nienhuis, A.W. 2002. High-level erythroid lineage-directed gene expression using globin gene regulatory elements after lentiviral vector-mediated gene transfer into primitive human and murine hematopoietic cells. Hum. Gene Ther. 13:2007–2016.

38. Persons, D.A., et al. 2003. Successful treatment of murine beta-thalassemia using in vivo selection of genetically modified, drug-resistant hematopoietic stem cells. Blood. 102:506–513.

39. Hong, S., et al. 2007. Functional analysis of various promoters in lentiviral vectors at different stages of

in vitro differentiation of mouse embryonic stem cells. Mol. Ther. 15:1630–1639.

40. Wang, L., Haas, D., Halene, S., and Kohn, D. 2003. Effects of the negative control region on expression from retroviral LTR. Mol. Ther. 7:438–439.

41. Challita, P.M., et al. 1995. Multiple modifications in cis elements of the long terminal repeat of ret-roviral vectors lead to increased expression and decreased DNA methylation in embryonic carci-noma cells. J. Virol. 69:748–755.

42. Robbins, P.B., et al. 1997. Increased probability of expression from modified retroviral vectors in embryonal stem cells and embryonal carcinoma cells. J. Virol. 71:9466–9474.

43. Halene, S., et al. 1999. Improved expression in hematopoietic and lymphoid cells in mice after transplantation of bone marrow transduced with a modified retroviral vector. Blood. 94:3349–3357.

44. Lois, C., Hong, E.J., Pease, S., Brown, E.J., and Bal-timore, D. 2002. Germline transmission and tis-sue-specific expression of transgenes delivered by lentiviral vectors. Science. 295:868–872.

45. Shepherd, B.E., et al. 2007. Hematopoietic stem-cell behavior in nonhuman primates. Blood. 110:1806–1813.

46. Abkowitz, J.L., Golinelli, D., Harrison, D.E., and Guttorp, P. 2000. In vivo kinetics of murine hemo-poietic stem cells. Blood. 96:3399–3405.

47. Zhou, S., et al. 2001. The ABC transporter Bcrp1/ABCG2 is expressed in a wide variety of stem cells and is a molecular determinant of the side-popula-tion phenotype. Nat. Med. 7:1028–1034.

48. Bunting, K.D. 2002. ABC transporters as pheno-typic markers and functional regulators of stem cells. Stem Cells. 20:11–20.

49. Moreb, J.S. 2008. Aldehyde dehydrogenase as a marker for stem cells. Curr. Stem Cell Res. Ther. 3:237–246.

50. Hess, D.A., et al. 2004. Functional characterization of highly purified human hematopoietic repopu-lating cells isolated according to aldehyde dehydro-genase activity. Blood. 104:1648–1655.

51. Sung, L.Y., et al. 2006. Differentiated cells are more efficient than adult stem cells for cloning by somat-ic cell nuclear transfer. Nat. Genet. 38:1323–1328.

52. Thorsteinsdottir, U., et al. 1997. Overexpression

of HOXA10 in murine hematopoietic cells per-turbs both myeloid and lymphoid differentiation and leads to acute myeloid leukemia. Mol. Cell Biol. 17:495–505.

53. Zhang, X.B., et al. 2008. High incidence of leukemia in large animals after stem cell gene therapy with a HOXB4-expressing retroviral vector. J. Clin. Invest. 118:1502–1510.

54. Larochelle, A., et al. 2006. AMD3100 mobilizes hematopoietic stem cells with long-term repopu-lating capacity in nonhuman primates. Blood. 107:3772–3778.

55. Donahue, R.E., et al. 1996. Peripheral blood CD34+ cells differ from bone marrow CD34+ cells in Thy-1 expression and cell cycle status in nonhuman pri-mates mobilized or non mobilized with granulo-cyte colony-stimulating factor and/or stem cell factor. Blood. 87:1644–1653.

56. Dunbar, C.E., et al. 1996. Improved retroviral gene transfer into murine and rhesus peripheral blood or bone marrow repopulating cells primed in vivo with stem cell factor and granulocyte colony-stimulating factor. Proc. Natl. Acad. Sci. U. S. A. 93:11871–11876.

57. Hanawa, H., et al. 2004. Efficient gene transfer into rhesus repopulating hematopoietic stem cells using a simian immunodeficiency virus-based len-tiviral vector system. Blood. 103:4062–4069.

58. Choi, U., et al. 2004. Nuclear-localizing O6-ben-zylguanine-resistant GFP-MGMT fusion protein as a novel in vivo selection marker. Exp. Hematol. 32:709–719.

59. Szymczak, A.L., et al. 2004. Correction of multi-gene deficiency in vivo using a single ‘self-cleaving’ 2A peptide-based retroviral vector. Nat. Biotechnol. 22:589–594.

60. Karolchik, D., et al. 2007. Comparative genomic analysis using the UCSC genome browser. Methods Mol. Biol. 395:17–34.

61. Kent, W.J. 2002. BLAT — the BLAST-like alignment tool. Genome Res. 12:656–664.

62. Futreal, P.A., et al. 2004. A census of human cancer genes. Nat. Rev. Cancer. 4:177–183.

63. Bradford, M.M. 1976. A rapid and sensitive method for the quantitation of microgram quantities of pro-tein utilizing the principle of protein-dye binding. Anal. Biochem. 72:248–254.

in vivo selection of hematopoietic progenitor cells and...

Documents